Social Network Analysis

import pandas as pd

# 미리 bookwarm library를 사용해서 구축해 놓은 인물 간 관계 데이터
book1 = pd.read_csv('harrypotter/network_HarryPotter1.csv', index_col=0)
book1.head()

import networkx as nx
import matplotlib.pyplot as plt

# Create an empty graph object
G = nx.Graph()

# Iterate through the DataFrame & add edges
for _, edge in book1.iterrows():
    G.add_edge(edge['source'], edge['target'], weight=edge['value'])

nx.draw_networkx(G)
plt.axis('off')  # turn off axis 
plt.show()

# degree centrality 계산
deg_cen = nx.degree_centrality(G)

# 각 node에 'centrality' attribute 넣어주기
for node in G.nodes():
    G.nodes[node]['centrality'] = deg_cen[node]


# centrality 값의 크기를 색으로 구분해 circos plot 그리기
import nxviz as nv

c = nv.CircosPlot(G, node_labels=True, 
                  node_label_layout="rotation",   # node끼리 서로 가리지 않도록 label를 rotate
                  node_color='centrality',   # centrality 값을 기반으로 색을 구분
                  node_order='centrality',   # centrality 값 순서대로 정렬
                  figsize=(12,12))
c.draw()

# 네트워크 G 안에서 특정 노드(name) 사이의 weight가 큰 neighbor 순서대로 10명 출력하는 함수
def find_ten_closest(G, name):  
    neighbors_list = list(G.neighbors(f"{name}"))
    
    temp_dict = {}
    for n in neighbors_list:
        weight = G.edges[f"{name}", n]['weight']
        temp_dict[n] = weight
    
    sorted_list = sorted(temp_dict.items(), key=(lambda x:x[1]), reverse=True)
    for n, w in sorted_list[0:10]:  # 1~10위
        print(f'{n}:  {w}')


# harry와 가장 가까운 인물 top10과 weight 출력 - 1권 기준
find_ten_closest(G, 'harry')

import pandas as pd
import networkx as nx

# 미리 bookwarm library를 사용해서 만들어 둔 인물 간 관계 데이터: 1권~7권
book1 = pd.read_csv('harrypotter/network_HarryPotter1.csv', index_col=0)
book2 = pd.read_csv('harrypotter/network_HarryPotter2.csv', index_col=0)
book3 = pd.read_csv('harrypotter/network_HarryPotter3.csv', index_col=0)
book4 = pd.read_csv('harrypotter/network_HarryPotter4.csv', index_col=0)
book5 = pd.read_csv('harrypotter/network_HarryPotter5.csv', index_col=0)
book6 = pd.read_csv('harrypotter/network_HarryPotter6.csv', index_col=0)
book7 = pd.read_csv('harrypotter/network_HarryPotter7.csv', index_col=0)


# 각 book으로부터 네트워크를 만들어 G_books list에 append
books = [book1, book2, book3, book4, book5, book6, book7]
G_books = []

for book in books:
    G_book = nx.Graph()
    for _, edge in book.iterrows():
        G_book.add_edge(edge['source'], edge['target'], weight=edge['value'])
    G_books.append(G_book)

# Creating a list of degree centrality of all the books
evol = [nx.degree_centrality(G) for G in G_books]
 
# Creating a DataFrame from the list of degree centralities in all the books
index = ['Book1', 'Book2', 'Book3', 'Book4', 'Book5', 'Book6', 'Book7']
degree_evol_df = pd.DataFrame.from_records(evol, index=index).fillna(0) # 모든 책에 모든 캐릭터가 있는 게 아니므로, N/A는 0으로 채워줌

# 각 책의 중요도 top 5 인물 선택
set_of_char = set()
for i in index:   # 1~7권
    set_of_char |= set(list(degree_evol_df.T[i].sort_values(ascending=False)[0:5].index)) 
            # |=는 왼쪽과 오른쪽의 합집합을 왼쪽에 assign해주는 역할
list_of_char = list(set_of_char)

# Evolution of Top Characters 시각화
degree_evol_df[list_of_char].plot(figsize=(13, 7));

# Creating a list of betweenness centrality of all the books
evol = [nx.betweenness_centrality(G, weight='weight') for G in G_books]

# Making a DataFrame from the list
index = ['Book1', 'Book2', 'Book3', 'Book4', 'Book5', 'Book6', 'Book7']
betweenness_evol_df = pd.DataFrame.from_records(evol, index=index).fillna(0)

# 각 책의 중요도 top 5 인물 선택
set_of_char = set()
for i in index:   # 1~7권
    set_of_char |= set(list(betweenness_evol_df.T[i].sort_values(ascending=False)[0:5].index))  
list_of_char = list(set_of_char)

# Evolution of Top Characters 시각화
betweenness_evol_df[list_of_char].plot(figsize=(13, 7));

# Creating a list of eigenvector centrality of all the books
evol = [nx.eigenvector_centrality(G, weight='weight') for G in G_books]

# Making a DataFrame from the list
index = ['Book1', 'Book2', 'Book3', 'Book4', 'Book5', 'Book6', 'Book7']
eigenvector_evol_df = pd.DataFrame.from_records(evol, index=index).fillna(0)

# 각 책의 중요도 top 5 인물 선택
set_of_char = set()
for i in index:   # 1~7권
    set_of_char |= set(list(eigenvector_evol_df.T[i].sort_values(ascending=False)[0:5].index))
list_of_char = list(set_of_char)

# Evolution of Top Characters 시각화
eigenvector_evol_df[list_of_char].plot(figsize=(13, 7));

# Creating a list of pagerank of all the characters in all the books
evol = [nx.pagerank(G) for G in G_books]

# Making a DataFrame from the list
index = ['Book1', 'Book2', 'Book3', 'Book4', 'Book5', 'Book6', 'Book7']
pagerank_evol_df = pd.DataFrame.from_records(evol, index=index).fillna(0)

# 각 책의 중요도 top 5 인물 선택
set_of_char = set()
for i in index:   # 1~7권
    set_of_char |= set(list(pagerank_evol_df.T[i].sort_values(ascending=False)[0:5].index))
list_of_char = list(set_of_char)

# Evolution of Top Characters 시각화
pagerank_evol_df[list_of_char].plot(figsize=(13, 7));